Re: [ndctl PATCH v4 09/17] util/hexdump: Add a util helper to print a buffer in hex

[Dan Williams <dan.j.williams@intel.com>]

On Thu, Oct 14, 2021 at 1:33 PM Verma, Vishal L
<vishal.l.verma@intel.com> wrote:
>
> On Thu, 2021-10-14 at 09:48 -0700, Dan Williams wrote:
> > On Thu, Oct 7, 2021 at 1:22 AM Vishal Verma <vishal.l.verma@intel.com> wrote:
> > >
> > > In preparation for tests that may need to set, retrieve, and display
> > > opaque data, add a hexdump function in util/
> > >
> > > Cc: Ben Widawsky <ben.widawsky@intel.com>
> > > Cc: Dan Williams <dan.j.williams@intel.com>
> > > Signed-off-by: Vishal Verma <vishal.l.verma@intel.com>
> > > ---
> > >  util/hexdump.h |  8 ++++++++
> > >  util/hexdump.c | 53 ++++++++++++++++++++++++++++++++++++++++++++++++++
> >
> > If this is just for tests shouldn't it go in tests/ with the other
> > common test helpers? If it stays in util/ I would kind of expect it to
> > use the log infrastructure, no?
>
> Agreed on using the log infra. I was originally using it in the old
> test stuff, but right now there's no user for it.. However having
> something like this was nice when developing the early cmd submission
> stuff. Do you think it would be good to always do a hexdump with dbg()
> when under --verbose for evert cxl_cmd_submit? (and maybe even add it
> for ndctl_cmd_submit later too) Or is that too noisy?

It sounds good as an extra-verbose debug option. At least it would be
more personal preference that -v does not get any more noisy by
default and require -vv to get hexdumps.

> If we want to do that then it makes sense to redo with the logging api,
> else maybe jsut drop this until we have a real in-tree user?

That's always ok in my book.